jectures. The results are quite unambiguous on this particularly in periods when the ceiling rate on time score deposits was increased. These periods exhibit rela- Additional information is supplied by Table I. For tively large contributions to the growth rate of bank each postwar cycle beginning with the downswing credit emanating from the time deposit substitution of 1948-49, the average annual growth rate of the mechanism money stock was computed. The expression M=mB A regression analysis (Table II)of th was then used to compute the contribution to the form equations derived from the Gramley-Chase average growth rate of money from three distinct model confirms the central role of the monetary base sources:(i)the behavior of monetary authorities(i.e, in the money supply process. Estimates of the regres e monet ry base and reserve requirement ratios), sion coefficient relating money to income are highly and the public's currency behavior, (ii) the time de- unstable among different sample periods, relative to posit substitution process, and,(iii) the variations in the coefficient relating money to the monetary base the excess reserve and borrowing ratios of commercial Furthermore, estimates of regression coefficients re- banks(Wicksell-Keynes mechanism lating money to income occur in some periods with TABLE II: A Comparison of Alternative Contributions to the Average essions of the Annual Growth Rate of the Money Stock and Bank Credit On the Monetary Base and G National Product* Rank Correlation Regression Coefficients For Bank Monetary Base Gross National Product Contribut wicksell- Keynes mechanism 04 1002 (3.12 growth rate of the money stock and bank credit with three different contributing 11/53 07 1.02 over all postwar half-cycles clearly support the con- 1 /57 clusion of the previous analysis that cyclical move- 1/60 11.76 11.81 【5.10) ments in the money stock are dominated by Federal 2.76 e actions Table I also presents the results of a similar . The monetary base was quarterly examination bearing on causes of movements in bank first entey iness, and partial corre lation coefficients oncoefficient.i-stotisies credit. The reader should note the radical difference ore in parent, o co in the observed patterns of correlation coefficients. signs which contradict the proposition of Gramley The behavior of monetary authorities, supplemented Chase and Cacy, or exhibit a very small statistical by the publics currency behavior, does not appear to significance. These diverse patterns of coefficients do dominate the behavior of bank credit. The three not occur for the estimates of coefficients relating sources contributing to the growth rate of money all money and the monetary base. It is also noteworthy exerted influences of similar order on bank credit. It that the average growth rate of the monetary base appears that bank credit is comparatively less exposed (adjusted for changes in reserve requirement ratios) to the push of Federal Reserve actions than was the over the upswings, exceeds without exception the money stock. On the other hand, the money stock is average growth rate of adjacent downswings. Thi less sensitive than bank credit to the time-deposit observation is not compatible with the contention substitution mechanism emphasized by gramley- made by Gramley-Chase that policy is countercyclical Chase, and the Wicksell-Keynes mechanism suggested Additional information is supplied by Table IIl by Cacy. Most astonishing, however, is the negatie which presents some results of a spectral analysis association between the average growth rate of bank credit and the Wicksell-Keynes mechanism em bearing on the monetary base and its sources. Spectral analysis is a statistical procedure for decomposing a time series into seasonal, cyclical, and trend It should also be noted that the average growth movements. After such an analysis was conducted ate of money conforms very clearly to the business the monetary base and its sources, a form of correla cycle. Such conformity does not hold for bank credit tion analysis was run between movements in the over the postwar half-cycles. This blurring occurred monetary base and movements in its various sources Page 17jectures. The results are quite unambiguous on this score. Additional information is supplied by Table I. For each postwar cycle beginning with the downswing of 1948-49, the average annual growth rate of the money stock was computed. The expression M mB was then used to compute the contribution to the average growth rate of money from three distinct sources: (i) the behavior of monetary authorities (i.e., the monetary base and reserve requirement ratios), and the public’s currency behavior, (ii) the time de￾posit substitution process, and, (iii) the variations in the excess reserve and borrowing ratios of commercial banks (Wicksell-Keynes mechanism). TABLE I: A Comparison of Alternative Contributions to the Average Annual Growth Rote of the Money Stock and Bank Credit Rank Correlations Contribution made by~ Public’s currency and authorities’ behavior Time deposit substitution mechanism Wicksell-Keynes mechanism Remarks The figures listed state the rank correlation between the average growth rate of the money stock and bank credit with three different contributing sources. The rank correlations between each contribution, and the average growth rate of the money stock over all postwar half-cycles clearly support the con￾clusion of the previous analysis that cyclical move￾ments in the money stock are dominated by Federal Reserve actions. Table I also presents the results of a similar examination bearing on causes of movements in bank credit. The reader should note the radical difference in the observed patterns of correlation coefficients. The behavior of monetary authorities, supplemented by the public’s currency behavior, does not appear to dominate the behavior of bank credit. The three sources contributing to the growth rate of money all exerted influences of similar order on bank credit. It appears that bank credit is comparatively less exposed to the push of Federal Reserve actions than was the money stock. On the other hand, the money stock is less sensitive than bank credit to the time-deposit substitution mechanism emphasized by Gramley￾Chase, and the Wicksell-Keynes mechanism suggested by Cacy. Most astonishing, however, is the negative association between the average growth rate of bank credit and the Wicksell-Keynes mechanism em￾phasized by Cacy. It should also be noted that the average growth rate of money conforms very clearly to the business cycle. Such conformity does not hold for bank credit over the postwar half-cycles. This blurring occurred particularly in periods when the ceiling rate on time deposits was increased. These periods exhibit rela￾tively large contributions to the growth rate of bank credit emanating from the time deposit substitution mechanism. A regression analysis (Table II) of the reduced form equations derived from the Gramley-Chase model confirms the central role of the monetary base in the money supply process. Estimates of the regres￾sion coefficient relating money to income are highly unstable among different sample periods, relative to the coefficient relating money to the monetary base. Furthermore, estimates of regression coefficients re￾lating money to income occur in some periods with TABLE II: Regressions of the Money Supply On the Monetary Base and Gross National Product* Moneta ry Base Gross National Product First Log First First Log First Differences Differences Differences Differences 2.03 .77 .04 .11 9.80) (10.02) (3.12) (3.39) .92 .93 .62 .65 ‘The monetary base was odiusted for res erye requirement changes and shift, in deposits. All data are quarterly averages of neasonally adiusted figures. The tirst entry in a column for each cycle is the regression coefficient. I-statistics are it pareetheses. and partial correlation coefficients are below the t-stottstics. signs which contradict the proposition of Gramley￾Chase and Cacy, or exhibit a very small statistical significance. These diverse patterns of coefficients do not occur for the estimates of coefficieilts relating money and the monetary base. It is also noteworthy that the average growth rate of the monetary base (adjusted for changes in reserve requirement ratios), over the upswings, exceeds without exception the average growth rate of adjacent downswings. This observation is not compatible with the contention made by Gramley-Chase that policy is countercyclical. Additional information is supplied by Table III, which presents some results of a spectral analysis bearing on the monetary base and its sources. Spectral analysis is a statistical procedure for decomposing a tune series into seasonal, cyclical, and trend movements. After such an analysis was conducted on the monetary base and its sources, a form of correla￾tion analysis was run between movements in the monetary base and movements in its various sources. Money 905 .048 .143 Bank Credit .333 .381 —-333 Regression Coefficients For: Cycle lV/48 to lI/S 3 11/53 to Ill/S 7 111/57 to 11/60 11/60 to Ill/os 1 .75 1.89) .44 4-59 (11.76) .97 2.76 7.56) .87 .63 1.96) .45 1.66 (11.81) .97 1.08 8.54) .89 .02 (1.02) .26 .06 (5.10) .86 —.01 (‘.33) -.08 .07 (1.23) .30 .19 (5.34) .67 -.03 (.27) Page 17